The present invention relates to text analytics, and more specifically, to disambiguating words within sentences. Text analytics relies on properly determining the use of words within a sentence, for example, determining the subject, action, and target of a sentence. One challenge in making such determinations is that some words have multiple meanings In addition, words may have different meanings depending on how the words are used in a sentence. For example, in the sentence “St. Paul loves you.” St. Paul could mean either a location (for example, St. Paul, Minn.) or a person. Even within the same paragraph, the same words can have different meanings.
Conventional text analytics methodologies lack means for determining the meaning of a word within a paragraph. However, deriving the author's intent for the use of the word is critical for extracting meaningful information from unstructured text.
Another example is “Charlotte is cool. Let's go there this weekend. Charlotte has a nice museum.” It would be helpful to know that the author of the text is referring to a city named Charlotte, and not to a person named Charlotte. Thus, there is a need for improved mechanisms for disambiguating words within text segments.